Genomic sequence variation within enhancers and promoters can have a significant impact on the cellular state and phenotype. However, sifting through the millions of candidate variants in a personal genome or a cancer genome, to identify those that impact cis-regulatory function, remains a major challenge. Interpretation of noncoding genome variation benefits from explainable artificial intelligence to predict and interpret the impact of a mutation on gene regulation. Here we generate phased whole genomes with matched chromatin accessibility, histone modifications, and gene expression for 10 melanoma cell lines. We find that training a specialized deep learning model, called DeepMEL2, on melanoma chromatin accessibility data can capture the various regulatory programs of the melanocytic and mesenchymal-like melanoma cell states. This model outperforms motif-based variant scoring, as well as more generic deep learning models. We detect hundreds to thousands of allele-specific chromatin accessibility variants (ASCAVs) in each melanoma genome, of which 15%-20% can be explained by gains or losses of transcription factor binding sites. A considerable fraction of ASCAVs are caused by changes in AP-1 binding, as confirmed by matched ChIP-seq data to identify allele-specific binding of JUN and FOSL1. Finally, by augmenting the DeepMEL2 model with ChIP-seq data for GABPA, the TERT promoter mutation, as well as additional ETS motif gains, can be identified with high confidence. In conclusion, we present a new integrative genomics approach and a deep learning model to identify and interpret functional enhancer mutations with allelic imbalance of chromatin accessibility and gene expression.

L’esperance de vie des patients atteints de melanome a haut risque ne peut etre predite d’une faconfiable en se basant sur les analyses d’histopathologies de la lesion primitive et est souvent ajusteedurant la progression de la maladie. Notre etude vise a elargir nos observations initiales au niveaudes metastases cutanees et d’evaluer la valeur pronostique de tyrosinase related protein 1 (TYRP1)dans les metastases ganglionnaires des patients atteints de melanome de stades III et IV. TYRP1 estune enzyme melanosomale qui partage des similitudes structurelles avec la tyrosinase, l'enzyme clede la melanogenese.L’expression de l'ARNm de TYRP1 a ete quantifiee dans 104 metastases ganglionnaires par PCRen temps reel et normalisee par rapport a l’expression de l’ARNm de S100B (marqueur reconnu dumelanome) pour corriger l’expression de TYRP1 suivant la charge tumorale de l’echantillon. Lerapport TYRP1/S100B a ete calcule et la mediane a ete utilisee en tant que valeur seuil. Ensuitenous avons etudie la relation entre les valeurs de TYRP1/S100B, le suivi clinique et lescaracteristiques histopathologiques de la tumeur primitive.Un rapport eleve de l’ARNm TYRP1/S100B correlait significativement avec une survie sansrecidive et une survie globale plus courtes, avec une epaisseur de Breslow plus elevee et avec lapresence d'une ulceration au niveau de la tumeur primitive. En outre, une expression elevee deTYRP1/S100B etait de meilleure valeur pronostique pour la survie globale que l'epaisseur deBreslow et l'ulceration des primitifs. De plus, cette expression est bien conservee au cours de laprogression de la maladie par rapport aux groupes de TYRP1 bas/eleve.Nous avons constate qu’une expression elevee de TYRP1/S100B dans les metastases de patientsatteints de melanome est associee a un resultat clinique defavorable et une survie courte. Menee surdes patients atteints d'un melanome a haut risque de recidive, cette premiere etude a suggere quel'ARNm de TYRP1 dans les metastases pourrait servir de biomarqueur pour affiner le pronosticinitial des patients surtout ceux ayant des lesions primitives de localisation inconnues ou nonevaluables et peut permettre une gestion differente des deux groupes de patients. Son expressionconservee au cours de la progression de la maladie est en faveur de son utilisation comme cibletherapeutique.En second lieu, en evaluant l’expression de la proteine TYRP1 par immunohistochimie dans lesmetastases cutanees et ganglionnaires, nous avons observe qu’elle n'etait pas detectee dans la moitie 7des tissus exprimant bel et bien l'ARNm correspondant et qu’elle, contrairement a l'ARNm, n’etaitpas associee a la survie.Des donnees recentes ont indique que le 3'-UTR de l’ARNm de TYRP1 contient trois sites deliaison putatifs de miR-155 dont deux presentant un polymorphisme d'un seul nucleotide (SNPs:rs683 et rs910) qui favorisent la degradation en cas d’hybridation miARN-ARNm parfaite del’ARNm ou non en cas d’hybridation imparfaite. Nous avons cherche a examiner si miR-155 peutaffecter l’expression de l’ARNm et de la proteine TYRP1 en fonction de ces SNPs. Tout d'abord,nous avons transfecte deux lignees de melanome ayant chacune l’une ou l’autre de l’allele (auniveau rs683 et rs910) avec differentes concentrations de pre-miR-155 et nous avons evaluel’expression du miR-155 et l’ARNm TYRP1 par PCR en temps reel ainsi que l’expression de laproteine TYRP1 par western blot. Nous avons constate qu’une surexpression de miR-155 a induitune degradation importante des ARNm TYRP1 et a perturbe sa traduction en proteine dans la ligneeavec le genotype “hybridation parfaite”. Ensuite, nous avons examine l'expression des ARNm etproteines de TYRP1, le niveau de miR-155 et les SNPs rs683 et rs910 dans 192 echantillons demetastases cutanees et ganglionnaires de melanome. Nous avons trouve que le groupe d'echantillonsavec le genotype “hybridation parfaite” etait significativement associe a un niveau de proteine deTYRP1 plus bas alors qu'aucune difference de niveau d’expression n'a ete trouvee pour l’ARNm deTYRP1 ou miR-155 entre les deux groupes de genotype, confirmant que les SNPs au niveau de 3’-UTR de TYRP1 peuvent specifiquement affecter l'expression de la proteine TYRP1. En outre, nousavons montre que l’ARNm de TYRP1 est inversement correle avec l’expression miR-155, mais pasavec la proteine TYRP1 dans le groupe parfaite, alors qu'il correle positivement avecla proteine mais pas avec miR-155 dans le groupe hybridation imparfaite ou la proteine correlaitinversement a la survie. Cela montre que les SNPs dans le 3'-UTR de l'ARNm TYRP1 affectent laregulation de l’ARNm par miR-155 et la traduction en proteine. Ces SNPs rendent la regulation del’ARNm et la proteine de TYRP1 independante de miR-155 et conferent une valeur pronostique ala proteine TYRP1

La depigmentation volontaire est une pratique bien connue en Afrique sub-saharienne. Elle se definit comme une pratique par laquelle une personne, de sa propre initiative, tente de diminuer la pigmentation melanique physiologique de sa propre peau. Les utilisateurs appliquent sur le corps, generalement sans surveillance medicale, de maniere soutenue et prolongee, des produits ou des melanges chimiques composes d’actifs depigmentants souvent d’une grande nocivite.Cette pratique est documentee dans plusieurs pays d’Afrique sub-saharienne (Senegal, Mali, Burkina Faso, Togo, Nigeria, ….), et sur d’autres continents. Face a l’absence de donnees chiffrees pour le Rwanda, nous avons realise une etude des pratiques de la depigmentation volontaire dans la capitale du pays, Kigali. Au Rwanda, certaines plantes etaient utilisees lors des grandes ceremonies comme le mariage, specialement par les femmes et les jeunes filles, pour eclaircir la peau. Une peau claire semble en fait un critere de beaute dans certaines traditions africaines. Nous avons donc realise une enquete ethnobotanique aupres de 61 tradipraticiens rwandais, afin de connaitre les plantes qui, avant l’arrivee de la cosmetique moderne, etaient utilisees pour « embellir » (eclaircir) la peau, afin de verifier si ces plantes pourraient interferer avec la production de la melanine. Notre enquete nous a permis de documenter 28 especes, dont cinq [Brillantaisia cicatricosa LINDAU; Chenopodium ugandae (Aellen) Aellen ; Dolichopentas longiflora Oliv.; Protea madiensis Oliv. subsp. Madiensis et Sesamum angolense Welw.] se sont distinguees par leur pourcentage de citation par les tradipraticiens. Ces dernieres ont fait objet de notre etude de laboratoire. Des extraits de polarite croissante, prepares a partir de ces cinq plantes, ont ete testes pour leur modulation de la melanogenese et de la tyrosinase (enzyme cle de la melanogenese) sur une serie de modeles: (i) sur la tyrosinase humaine dans les extraits totaux de melanocytes normaux; (ii) sur des melanocytes malins en culture (pour evaluer l’effet global des extraits de plante sur la melanogenese); (iii) sur la tyrosinase de champignon en solution et sur chromatoplaque de silice; et enfin (iv) sur l’activite tyrosine hydroxylase de l'enzyme. Deux extraits a l’acetate d’ethyle de Protea madiensis Oliv. et de Sesamum angolense Welw. ont ete selectionnes pour leur activite, respectivement inhibitrice et activatrice de la tyrosinase de champignon. Ces deux extraits ont ete soumis a une serie de fractionnements dans le but d’isoler et d’identifier des composes actifs. Trois composes ont ete isoles de Protea madiensis (2-tridecanone, acide oleique et β-sitosterol). La 2-tridecanone et l’acide oleique ont montre une inhibition de la tyrosinase de champignon sur chromatoplaque et de la tyrosinase humaine dans les extraits cellulaires. De plus, la 2-tridecanone a montre une inhibition de l’activite tyrosine hydroxylase. Le β-sitosterol n’a pas montre d’effet sur nos modeles mais il a deja ete isole dans d’autres etudes en tant qu'inhibiteur de la tyrosinase. De l’extrait a l’acetate d’ethyle de Sesamum angolense Welw., nous avons isole l’acide ursolique qui a montre une augmentation de l’activite de la tyrosinase de champignon sur chromatoplaque.L’enquete ethnobotanique nous a permis de constater que la flore rwandaise regorge de plantes aux vertus cosmetiques interessantes; celles-ci pourraient representer une alternative aux actifs depigmentants connus pour leurs nombreux effets secondaires mais neanmoins largement disponibles sur le marche rwandais. L’enquete realisee dans la ville de Kigali, nous a permis de constater que 27 % de notre population d’etude sont des utilisateurs conscients de produits depigmentants. Ce pourcentage nous semble fort eleve et des mesures devraient etre prises pour la sensibilisation et la conscientisation de la population quant aux risques encourus et a l’existence de medecines traditionnelles a visee depigmentante. Ces mesures devraient etre combinees avec la recherche de composes naturels dans l'espoir d'identifier des molecules actives et faiblement toxiques, voire atoxiques. L’etude de la modulation de la pigmentation par les extraits des cinq plantes selectionnees, nous a permis de confirmer l’information recue des tradipraticiens. Cette etude nous a egalement montre que ces extraits de plantes renferment des activateurs de la melanogenese, qui pourraient etre exploites pour le bronzage recherche par les sujets de peau claire.L’isolement et identification de molecules a partir des extraits de deux plantes, nous a permis de constater que notre methode de bioguidage fonctionne correctement; des mesures de dereplications devraient cependant etre prises pour eviter autant que possible de retomber sur des molecules deja connues./Voluntary depigmentation, well-known in sub-Saharan Africa, is defined as a practice by which a person, by his/her own initiative, attempts to reduce his/her skin physiological melanin pigmentation. Users apply on the body, usually without medical supervision, in a sustained and prolonged manner, depigmenting compounds, single or in mixtures.This quite harmful practice is documented in several sub-Saharan African countries (Senegal, Mali, Togo, Nigeria…) and in other continents. The absence of Rwandese data prompted us to conduct a study of the practices of voluntary depigmentation in the capital, Kigali.In Rwanda, some plants were used during important ceremonies like wedding (marriage) especially by women and girls to lighten their skin. Fair skin is actually considered as a beauty criterion in some African traditions.We conducted an ethnobotanical survey of 61 Rwandan traditional healers to identify the plants that were used before the introduction of modern cosmetics to beautify (lighten) the skin in order to check wether these plants could interfere with the production of melanin.Our survey allowed us to identify and collect 28 species, of which 5 were selected (retained) for their higher percentage of citation by traditional healers [Brillantaisia cicatricosa LINDAU; Chenopodium ugandae (Aellen) Aellen ; Dolichopentas longiflora Oliv.; Protea madiensis Oliv. subsp. madiensis and Sesamum angolense Welw.]. These five species have been used for our laboratory study.Extracts of increasing polarities were prepared from the five plants and tested for their ability to modulate melanogenesis and tyrosinase (the key enzyme of melanogenesis) in a series of models: (i) human tyrosinase in total extracts from normal melanocytes; (ii) malignant melanocytes in culture (in order to assess the global effect of plant extracts on melanogenesis); (iii) mushroom tyrosinase in solution and on TLC plate; and finally (iv) tyrosine hydroxylase activity of the enzyme.Two ethyl acetate extracts of Protea madiensis Oliv. and of Sesamum angolense Welw have been selected according to their respective inhibitory and activating effect on mushroom tyrosinase. These two extracts were fractionated to isolate and identify active compounds. Three compounds have been isolated from Protea madiensis (2-tridecanone, oleic acid and β-sitosterol). The 2-tridecanone and the oleic acid showed an inhibition of mushroom tyrosinase on TLC and human tyrosinase in cellular extracts. In addition, 2-tridecanone showed an inhibition of the tyrosine hydroxylase activity. β-sitosterol showed no effect on our models but has been identified, in other studies, as a tyrosinase inhibitor. From the ethyl acetate extract of Sesamum angolense, we isolated ursolic acid which increases the mushroom tyrosinase activity on TLC.The ethnobotanical survey allowed us to (state) notice that Rwandan flora contains plants that have interesting cosmetic properties and could be an alternative to the use of harmful depigmenting products which are sold on Rwandese markets.The survey conducted in Kigali city indicates that 27 % of surveyed persons are conscious users of depigmenting products. This percentage seems very high so that measures should be taken to raise awareness about the involved risks and of the existence of traditional medicines with such depigmenting effects. These measures should be accompanied (combined) with the search for natural compounds with depigmenting effect in the hope to identify actives that would be weakly or even non toxic at all.The study of the pigmentation modulation by five selected plant extracts allowed to confirm the information obtained from traditional healers. It also indicates that, apart from an inhibitory effect, some of our plant extracts also contain melanogenesis activators that could be further exploited for tanning, an aspiration of fair-skinned individuals.The isolation and identification of molecules from two plants extracts led us to conclude that our “bioguidance” method performs adequately. Nevertheless, some dereplication measures should be implemented to avoid spending time on isolating already known molecules.

Reference EPFL-CONF-171519View record in Web of Science Record created on 2011-12-16, modified on 2017-05-12

The search for more effective treatment strategies in melanoma led to many new innovative approaches aiming at different molecular targets. Chemotherapy still remains the most effective treatment and many efforts are put in order to improve targeting and delivery of the chemotherapeutic agents. Among these, peptide conjugates of anticancer drugs were designed to increase stability, cell penetration, specificity and accumulation in cancer cells. We as well as others evaluated such a conjugate, termed PSF (L-prolyl-m-L-sarcolysyl-L-p-fluorophenylalanine-ethylester) in terms of its cytotoxicity in vitro and in vivo using a human melanoma tumor as a model, its stability, transport, and metabolisation. By comparing the cytotoxicity of PSF and melphalan towards different cancer primary melanoma cell cultures, we noticed some interesting observations: PSF displayed the same toxicity pattern both in short (2h) and long term (24h) cell exposures whereas melphalan and m-sarcolysin needed long term exposure to reach the same toxicity. This could indicate that PSF very quickly penetrates the cells in accordance with what has been shown with red blood cells (RBCs). PSF has shown a much better and quicker penetration into the cells in vitro as compared to melphalan. In this present work, the cytotoxic effect of PSF was further evaluated in vivo using a standardized nude mice tumor model bearing a human melanoma. First, the acute toxicity in rats and mice and the maximum tolerated dose were determined. After a dose-escalation study one dose was singled out and tested as a single dose and as a fractionated dose. PSF was able to reach the tumor site and a dose-response relationship was observed. The IP administration of fractionated doses of PSF had significantly better effect on tumor growth inhibition, regression and regrowth than single dose administration and this without any evidence for general toxicity monitored by animal weight loss. We also compared the efficacy of PSF to its parent drug m-sarcolysin, melphalan and cyclophosphamide and observed that PSF was much more active than both melphalan and m-sarcolysin at the same molar doses.Body distribution of the 14C-labelled PSF revealed ratios of 2.4 and 1.5 compared to muscle tissue for the two melanoma tumors evaluated with no significant and stable accumulation in any vital organ. The amount of tracer was still high in the blood after 24 hours explaining the high radioactivity in the kidney and partly in the liver. Interestingly, the spleen had an unusual high radioactivity uptake reflecting the exceptional binding of the tracer to blood cells (BC), while the pancreas very high load was an indicator of protease-mediated specific delivery and strongly support our hypothesis elaborated on the basis of in vitro results. Our in vitro data point to a particular mechanism of action of PSF based on the transport of PSF through the body by the rapid binding to blood cells and the delivery at the tumor site by the subsequent release of its active metabolites due to cleavage by tumor-associated proteases.Concerning the binding of PSF to membranes and its transport the following observations were made: while PSF was stable in human plasma, it disappeared very quickly in whole blood along with the generation of a main metabolite: m-sarcolysin. The presence of BC membranes was required for both binding and generating the metabolites. Binding to natural or artificial membranes was achieved and only competition with melanoma cells or proteolytic enzymes such as dispase, led to the generation of active metabolites. The different metabolites were isolated using preparative LC and were then identified using Electrospray Ionisation Mass Spectrometry (ESI). Three metabolites, of which m-sarcolysin was the main one, were identified all bearing the chloroethyl alkylating group. Enzymatic catalysis was further supported by a set of experiments where the enzymatic activity was non-specifically and specifically inhibited. In order to look at the effect of extracellular matrix proteases on PSF, three representatives of ECM proteases were incubated with PSF: collagenase A had no effect, but both dispase and trypsine were able to process PSF. The following data indicate the higher processing of PSF in the presence of cells with a higher proteolytic activity and thus the delivery of the blood cell-bound PSF. When comparing BC with melanoma cells (MC), the latter showed a higher ability to bind and process PSF both by membrane-associated and most interestingly soluble proteases. A lot of families of enzymes are reported to be overexpressed by melanoma cells including: metalloproteases, cysteine cathepsins, serine proteases and aminopeptidases. All the melanoma cells and cell lines evaluated were able to generate PSF active metabolites. To identify the families of enzymes expressed on the membrane of melanoma cells that might be involved in the mechanism of action of PSF, we performed 2D-gel electrophoresis on their membrane extracts. The 2D-gels experiments revealed the presence of proteins compatible with enzymes known to be important in melanoma and further work is needed to identify the individual enzymes involved by using mass spectrometry and Western blotting. Both our in vitro and in vivo findings strongly suggest that not only melanoma tumor cells and tumor sites but other types of tumors as well may be targets for the toxic activity of PSF owing to their much higher load in proteolytic enzymes that are closely related to their invasive potential. The transport of PSF by the blood cells and the release of its metabolites at the tumor site result in a low amount of drug in its free soluble form within the blood and this may explain the relatively lower side-effects observed. PSF is thus expected to have a much better therapeutic index than conventional alkylating agents. This original mechanism of drug delivery may well be extended to other cancer and non-cancer drugs than alkylating agents.

Irofulven is currently in Phase 2 clinical trials against a wide variety of solid tumors and has demonstrated activity in ovarian, prostate, gastrointestinal, and non–small cell lung cancer. The objectives of this study were to determine its pharmacokinetics and route of excretion and to characterize its metabolites in human plasma and urine samples after a 30-min i.v. infusion at a dose of 0.55 mg/kg in patients with advanced solid tumors. Three patients were administered i.v. 100 Ci of [ 14 C]irofulven over a 30-min infusion on day 1 of cycle 1. Serial blood and plasma samples were drawn at 0 (before irofulven infusion) and up to 144 h after the start of infusion. Urine and fecal samples were collected for up to 144 h after the start of infusion. The mean urinary and fecal excretion of radioactivity up to 144 h were 71.2 and 2.9%, respectively, indicating renal excretion was the major route of elimination of [ 14 C]irofulven. The Cmax, AUC0-, and terminal half-life values for total radioactivity were 1130 ng-Eq/ml, 24,400 ng-Eq h/ml, and 116.5 h, respectively, and the corresponding values for irofulven were 82.7 ng/ml, 65.5 ng h/ml, and 0.3 h, respectively, suggesting that the total radioactivity in human plasma was a result of the metabolites. Twelve metabolites of irofulven were detected in human urine and plasma by electrospray ionization/tandem mass spectrometry. Among these metabolites, the cyclopropane ring-opened metabolite (M2) of irofulven was found, and seven others were proposed as glucuronide and glutathione conjugates.